How do you calculate the minimum detectable defect for a camera?

Divide the field of view by the camera's pixels across it to get the pixel size at the part, then multiply by the pixels a defect must span. 220 mm over 2,448 pixels is 0.0899 mm a pixel; at 3 pixels per defect the minimum defect is 0.270 mm, 0.030 mm inside a 0.3 mm specification, the Minimum Detectable Defect calculator's own example.

How many pixels should a defect span?

Your own validated figure for your algorithm, lighting and defect type: the number of pixels at which the inspection finds the defect reliably on real parts. The example uses 2 to 4; it is an input, not a recommendation.

What is the widest field of view column?

The widest view at which the station's camera still resolves the specified defect: specified defect x horizontal pixels / pixels per defect. In the example the barrel scratch station needs 40.8 mm or less with its 2,448 pixel camera, against the 45 mm it uses.

Does a station that resolves its defect always find it?

No. The workbook gives the sampling limit set by the pixel size. Contrast, lighting, focus, depth of field, lens resolution and motion blur all decide whether a defect is found, so a station within limits here still needs validating on real parts.

Which pixel count should I use, horizontal or vertical?

The one in the same direction as the field of view you type. For a defect that can lie either way, check the shorter sensor direction too, with its field of view.

Does the workbook use macros?

No. It is a plain .xlsx with ordinary formulas, tables and native charts. The Verify sheet re-checks the formulas against fixed test stations and shows Self-check: PASS.

Quality and Inspection

Machine Vision Minimum Defect Size Excel Template

Minimum defect = field of view ÷ horizontal pixels × your validated pixels per defect. Give every inspection station a row and see whether its camera and field of view can resolve the defect it must catch, the headroom, the widest view that would, and every camera at the station you pick.

Free, and a plain .xlsx with no macros to enable. A CSV with live formulas and a printed how-to guide come with it.

Overview

This machine vision minimum defect size template is a working Excel workbook for vision system engineers, quality and test engineers and automation integrators. Each inspection station is one row with its camera, horizontal field of view, the pixels a defect must span for your inspection to find it, and the defect size the station must find. The workbook works out the pixel size at the part, the smallest defect the station can sample, the headroom against the specification and the widest field of view that would still resolve it. It ships with seven stations on a fictional medical device line, so every tile and chart is filled in when you open it.

In the example the barrel scratch station uses a 2,448 pixel camera over a 45 mm view: 0.0184 mm a pixel, so at 3 pixels per defect the smallest scratch it can sample is 0.055 mm, larger than the 0.05 mm specification. With that camera the view would have to narrow to 40.8 mm; the what-if shows the 4,096 pixel camera at 45 mm resolving 0.033 mm. The other six stations resolve their defects, the connector pin station with only 0.005 mm to spare.

This is the sampling limit only: optics, lighting, contrast, focus and motion blur decide what a station really finds, so validate every inspection on real parts. The download is a plain .xlsx with no macros. A CSV with live formulas is also included for other tools; the dashboard, charts, dropdowns and self-check are in the .xlsx.

What this template includes

Setup sheet with your line name and a camera list carrying each camera's horizontal pixel count
Stations table, one row per inspection: camera, horizontal field of view, validated pixels per defect and specified defect size, with a Check column
Per station: object-space pixel size, sampling-limited minimum defect, pixels across the specified defect, headroom, a status in words and the widest field of view that still resolves the defect
Dashboard with four tiles (stations, stations not resolving their defect, smallest headroom, finest pixel size), minimum defect against specification, pixels across the defect by station, a what-if of every camera at the station you pick, and the station table
Method sheet with every formula, and a Verify sheet that re-checks the formulas against an independent reference and the Minimum Detectable Defect calculator (Self-check: PASS)
CSV with live formulas for the Stations table; the dashboard, charts, dropdowns and self-check are in the .xlsx

Suggested use case

Use it to specify a camera and field of view for a new inspection, to audit the stations on an existing line against their specifications, or to see what a narrower view or a higher-resolution camera would change.

How to use it

On Setup, type the line name and list your cameras with their horizontal pixel count from the data sheet.
On Stations, give each inspection a row: camera, horizontal field of view, your validated pixels per defect and the specified defect size.
Fix anything the Check column names.
Open the Dashboard: stations that do not resolve their defect, the headroom of each, and every camera at the station you pick.

Frequently Asked Questions

How do you calculate the minimum detectable defect for a camera?: Divide the field of view by the camera's pixels across it to get the pixel size at the part, then multiply by the pixels a defect must span. 220 mm over 2,448 pixels is 0.0899 mm a pixel; at 3 pixels per defect the minimum defect is 0.270 mm, 0.030 mm inside a 0.3 mm specification, the Minimum Detectable Defect calculator's own example.
How many pixels should a defect span?: Your own validated figure for your algorithm, lighting and defect type: the number of pixels at which the inspection finds the defect reliably on real parts. The example uses 2 to 4; it is an input, not a recommendation.
What is the widest field of view column?: The widest view at which the station's camera still resolves the specified defect: specified defect x horizontal pixels / pixels per defect. In the example the barrel scratch station needs 40.8 mm or less with its 2,448 pixel camera, against the 45 mm it uses.
Does a station that resolves its defect always find it?: No. The workbook gives the sampling limit set by the pixel size. Contrast, lighting, focus, depth of field, lens resolution and motion blur all decide whether a defect is found, so a station within limits here still needs validating on real parts.
Which pixel count should I use, horizontal or vertical?: The one in the same direction as the field of view you type. For a defect that can lie either way, check the shorter sensor direction too, with its field of view.
Does the workbook use macros?: No. It is a plain .xlsx with ordinary formulas, tables and native charts. The Verify sheet re-checks the formulas against fixed test stations and shows Self-check: PASS.